Chore: Parse quoted text in incoming emails (#883) (#884)

* Chore: Parse quoted text in incoming emails (#883)
* Parsed the quoted text and replies in incoming emails and store them separately
* Did this parsing for plain text and html part of emails
* In the chat window, we will only show the parsed reply alone

* Conversation mailbox test fixes (#883)

app/mailboxes/conversation_mailbox.rb

@@ -21,7 +21,7 @@ class ConversationMailbox < ApplicationMailbox
     @message = @conversation.messages.create(
       account_id: @conversation.account_id,
       contact_id: @conversation.contact_id,
-      content: processed_mail.content,
+      content: processed_mail.text_content[:reply],
       inbox_id: @conversation.inbox_id,
       message_type: 'incoming',
       content_type: 'incoming_email',
@@ -71,6 +71,6 @@ class ConversationMailbox < ApplicationMailbox
   end
 
   def decorate_mail
-    @processed_mail = MailPresenter.new(mail)
+    @processed_mail = MailPresenter.new(mail, @conversation.account)
   end
 end

app/presenters/mail_presenter.rb

@@ -1,21 +1,32 @@
 class MailPresenter < SimpleDelegator
   attr_accessor :mail
 
-  def initialize(mail)
+  def initialize(mail, account = nil)
     super(mail)
     @mail = mail
+    @account = account
   end
 
   def subject
     encode_to_unicode(@mail.subject)
   end
 
-  def content
-    return @decoded_content if @decoded_content
+  def text_content
+    @decoded_text_content ||= encode_to_unicode(text_part&.body&.decoded || '')
+    @text_content ||= {
+      full: @decoded_text_content,
+      reply: extract_reply(@decoded_text_content)[:reply],
+      quoted: extract_reply(@decoded_text_content)[:quoted_text]
+    }
+  end
 
-    @decoded_content = parts.present? ? parts[0].body.decoded : decoded
-    @decoded_content = encode_to_unicode(@decoded_content)
-    @decoded_content
+  def html_content
+    @decoded_html_content ||= encode_to_unicode(html_part&.body&.decoded || '')
+    @html_content ||= {
+      full: @decoded_html_content,
+      reply: extract_reply(@decoded_html_content)[:reply],
+      quoted: extract_reply(@decoded_html_content)[:quoted_text]
+    }
   end
 
   def attachments
@@ -36,7 +47,8 @@ class MailPresenter < SimpleDelegator
 
   def serialized_data
     {
-      content: content,
+      text_content: text_content,
+      html_content: html_content,
       number_of_attachments: number_of_attachments,
       subject: subject,
       date: date,
@@ -56,4 +68,35 @@ class MailPresenter < SimpleDelegator
     current_encoding = str.encoding.name
     str.encode(current_encoding, 'UTF-8', invalid: :replace, undef: :replace, replace: '?')
   end
+
+  def extract_reply(content)
+    @regex_arr ||= quoted_text_regexes
+
+    content_length = content.length
+    # calculates the matching regex closest to top of page
+    index = @regex_arr.inject(content_length) do |min, regex|
+      [(content.index(regex) || content_length), min].min
+    end
+
+    {
+      reply: content[0..(index - 1)].strip,
+      quoted_text: content[index..-1].strip
+    }
+  end
+
+  def quoted_text_regexes
+    sender_agnostic_regexes = [
+      Regexp.new("^.*On.*(\n)?wrote:$", Regexp::IGNORECASE),
+      Regexp.new("-+original\s+message-+\s*$", Regexp::IGNORECASE),
+      Regexp.new("from:\s*$", Regexp::IGNORECASE)
+    ]
+    return sender_agnostic_regexes if @account.nil? || @account.support_email.blank?
+
+    [
+      Regexp.new("From:\s*" + Regexp.escape(@account.support_email), Regexp::IGNORECASE),
+      Regexp.new('<' + Regexp.escape(@account.support_email) + '>', Regexp::IGNORECASE),
+      Regexp.new(Regexp.escape(@account.support_email) + "\s+wrote:", Regexp::IGNORECASE),
+      Regexp.new('On(.*)' + Regexp.escape(@account.support_email) + '(.*)wrote:', Regexp::IGNORECASE)
+    ] + sender_agnostic_regexes
+  end
 end

spec/mailboxes/conversation_mailbox_spec.rb

@@ -8,7 +8,7 @@ RSpec.describe ConversationMailbox, type: :mailbox do
     let(:reply_mail) { create_inbound_email_from_fixture('reply.eml') }
     let(:conversation) { create(:conversation, assignee: agent) }
     let(:described_subject) { described_class.receive reply_mail }
-    let(:serialized_attributes) { %w[content number_of_attachments subject date to from in_reply_to cc bcc message_id] }
+    let(:serialized_attributes) { %w[text_content html_content number_of_attachments subject date to from in_reply_to cc bcc message_id] }
 
     before do
       # this UUID is hardcoded in the reply.eml, that's why we are updating this
@@ -19,7 +19,7 @@ RSpec.describe ConversationMailbox, type: :mailbox do
     end
 
     it 'add the mail content as new message on the conversation' do
-      expect(conversation.messages.last.content).to eq("Let's talk about these images:\r\n\r\n")
+      expect(conversation.messages.last.content).to eq("Let's talk about these images:")
     end
 
     it 'add the attachments' do

spec/presenters/mail_presenter_spec.rb

@@ -8,7 +8,7 @@ RSpec.describe MailPresenter do
 
     it 'give utf8 encoded content' do
       expect(decorated_mail.subject).to eq("Discussion: Let's debate these attachments")
-      expect(decorated_mail.content).to eq("Let's talk about these images:\r\n\r\n")
+      expect(decorated_mail.text_content[:full]).to eq("Let's talk about these images:\r\n\r\n")
     end
 
     it 'give decoded blob attachments' do
@@ -24,7 +24,10 @@ RSpec.describe MailPresenter do
 
     it 'give the serialized data of the email to be stored in the message' do
       data = decorated_mail.serialized_data
-      expect(data.keys).to eq([:content, :number_of_attachments, :subject, :date, :to, :from, :in_reply_to, :cc, :bcc, :message_id])
+      expect(data.keys).to eq([
+                                :text_content, :html_content, :number_of_attachments, :subject, :date, :to,
+                                :from, :in_reply_to, :cc, :bcc, :message_id
+                              ])
       expect(data[:subject]).to eq(decorated_mail.subject)
       expect(data[:date].to_s).to eq('2020-04-20T04:20:20-04:00')
       expect(data[:message_id]).to eq(mail.message_id)